County-level intensity of carbon emissions from crop farming in China during 2000-2019.

Agriculture is an important contributor to global carbon emissions. With the implementation of the Sustainable Development Goals of the United Nations and China's carbon neutral strategy, accurate estimation of carbon emissions from crop farming is essential to reduce agricultural carbon emissions and promote sustainable food production systems in China. However, previous long-term time series estimates in China have mainly focused on the national and provincial levels, which are insufficient to characterize regional heterogeneity. Here, we selected the county-level administrative district as the basic geographical unit and then generated a county-level dataset on the intensity of carbon emissions from crop farming in China during 2000-2019, using random forest regression with multi-source data. This dataset can be used to delineate spatio-temporal changes in carbon emissions from crop farming in China, providing an important basis for decision makers and researchers to design agricultural carbon reduction strategies in China.

Unveiling diverse coordination-defined electronic structures of reconstructed anatase TiO2(001)-(1 × 4) surface.

Transition metal oxides (TMOs) exhibit fascinating physicochemical properties, which originate from the diverse coordination structures between the transition metal and oxygen atoms. Accurate determination of such structure-property relationships of TMOs requires to correlate structural and electronic properties by capturing the global parameters with high resolution in energy, real, and momentum spaces, but it is still challenging. Herein, we report the determination of characteristic electronic structures from diverse coordination environments on the prototypical anatase-TiO2(001) with (1 × 4) reconstruction, using high-resolution angle-resolved photoemission spectroscopy and scanning tunneling microscopy/atomic force microscopy, in combination with density functional theory calculation. We unveil that the shifted positions of O 2s and 2p levels and the gap-state Ti 3p levels can sensitively characterize the O and Ti coordination environments in the (1 × 4) reconstructed surface, which show distinguishable features from those in bulk. Our findings provide a paradigm to interrogate the intricate reconstruction-relevant properties in many other TMO surfaces.

Hexagonal image segmentation on spatially resolved transcriptomics.

Spatial transcriptomics is a rapidly evolving field that enables researchers to capture comprehensive molecular profiles while preserving information about the physical locations. One major challenge in this research area involves the identification of spatial domains, which are distinct regions characterized by unique gene expression patterns. However, current unsupervised methods have struggled to perform well in this regard due to the presence of high levels of noise and dropout events in spatial transcriptomic profiles. In this paper, we propose a novel hexagonal Convolutional Neural Network (hexCNN) for hexagonal image segmentation on spatially resolved transcriptomics. To address the problem of noise and dropout occurrences within spatial transcriptomics data, we first extend an unsupervised algorithm to a supervised learning method that can identify useful features and reduce noise hindrance. Then, inspired by the classical convolution in convolutional neural networks (CNNs), we designed a regular hexagonal convolution to compensate for the missing gene expression patterns from adjacent spots. We evaluated the performance of hexCNN by applying it to the DLPFC dataset. The results show that hexCNN achieves a classification accuracy of 86.8% and an average Rand index (ARI) of 77.1% (1.4% and 2.5% higher than those of GNNs). The results also demonstrate that hexCNN is capable of removing the noise caused by batch effect while preserving the biological signal differences.

Reassessment of growth-climate relations indicates the potential for decline across Eurasian boreal larch forests.

Larch, a widely distributed tree in boreal Eurasia, is experiencing rapid warming across much of its distribution. A comprehensive assessment of growth on warming is needed to comprehend the potential impact of climate change. Most studies, relying on rigid calendar-based temperature series, have detected monotonic responses at the margins of boreal Eurasia, but not across the region. Here, we developed a method for constructing temporally flexible and physiologically relevant temperature series to reassess growth-temperature relations of larch across boreal Eurasia. Our method appears more effective in assessing the impact of warming on growth than previous methods. Our approach indicates widespread and spatially heterogeneous growth-temperature responses that are driven by local climate. Models quantifying these results project that the negative responses of growth to temperature will spread northward and upward throughout this century. If true, the risks of warming to boreal Eurasia could be more widespread than conveyed from previous works.

Two-Dimensional Iron Silicide (FeSix) Alloys with Above-Room-Temperature Ferromagnetism.

Two-dimensional (2D) materials with intrinsic room-temperature ferromagnetism have gathered tremendous interest as promising candidates for next-generation spintronics. Here, on the basis of first-principles calculations, we report a family of stable 2D iron silicide (FeSix) alloys via dimensional reduction of their bulk counterparts. Our results demonstrate that 2D Fe4Si2-hex, Fe4Si2-orth, Fe3Si2, and FeSi2 nanosheets are lattice-dynamically and thermally stable, confirmed by the calculated phonon spectra and Born-Oppenheimer dynamic simulation up to 1000 K. 2D FeSix nanosheets are ferromagnetic metals with estimated Curie temperatures ranging from 547 to 971 K due to strong direct exchange interaction between Fe sites. In addition, the electronic properties of 2D FeSix alloys can be maintained on silicon substrates, providing an ideal platform for spintronics applications in the nanoscale.

Prediction of protein-ligand binding affinity with deep learning.

The prediction of binding affinities between target proteins and small molecule drugs is essential for speeding up the drug research and design process. To attain precise and effective affinity prediction, computer-aided methods are employed in the drug discovery pipeline. In the last decade, a variety of computational methods has been developed, with deep learning being the most commonly used approach. We have gathered several deep learning methods and classified them into convolutional neural networks (CNNs), graph neural networks (GNNs), and Transformers for analysis and discussion. Initially, we conducted an analysis of the different deep learning methods, focusing on their feature construction and model architecture. We discussed the advantages and disadvantages of each model. Subsequently, we conducted experiments using four deep learning methods on the PDBbind v.2016 core set. We evaluated their prediction capabilities in various affinity intervals and statistically and visually analyzed the samples of correct and incorrect predictions for each model. Through visual analysis, we attempted to combine the strengths of the four models to improve the Root Mean Square Error (RMSE) of predicted affinities by 1.6% (reducing the absolute value to 1.101) and the Pearson Correlation Coefficient (R) by 2.9% (increasing the absolute value to 0.894) compared to the current state-of-the-art method. Lastly, we discussed the challenges faced by current deep learning methods in affinity prediction and proposed potential solutions to address these issues.

ContactLib-ATT: a structure-based search engine for homologous proteins.

General-purpose protein structure embedding can be used for many important protein biology tasks, such as protein design, drug design and binding affinity prediction. Recent researches have shown that attention-based encoder layers are more suitable to learn high-level features. Based on this key observation, we propose a two-level general-purpose protein structure embedding neural network, called ContactLib-ATT. On local embedding level, a biologically more meaningful contact context is introduced. On global embedding level, attention-based encoder layers are employed for better global representation learning. Our general-purpose protein structure embedding framework is trained and tested on the SCOP40 2.07 dataset. As a result, ContactLib-ATT achieves a SCOP superfamily classification accuracy of 82.4% (i.e., 6.7% higher than state-of-the-art method). On the same dataset, ContactLib-ATT is used to simulate a structure-based search engine for remote homologous proteins, and our top-10 candidate list contains at least one remote homolog with a probability of 91.9%.

Hydrogen-Bond Network Promotes Water Splitting on the TiO2 Surface.

Breaking the strong covalent O-H bond of an isolated H2O molecule is difficult, but it can be largely facilitated when the H2O molecule is connected with others through hydrogen-bonding. How a hydrogen-bond network forms and performs becomes crucial for water splitting in natural photosynthesis and artificial photocatalysis and is awaiting a microscopic and spectroscopic understanding at the molecular level. At the prototypical photocatalytic H2O/anatase-TiO2(001)-(1×4) interface, we report the hydrogen-bond network can promote the coupled proton and hole transfer for water splitting. The formation of a hydrogen-bond network is controlled by precisely tuning the coverage of water to above one monolayer. Under ultraviolet (UV) light irradiation, the hydrogen-bond network opens a cascaded channel for the transfer of a photoexcited hole, concomitant with the release of the proton to form surface hydroxyl groups. The yielded hydroxyl groups provide excess electrons to the TiO2 surface, causing the reduction of Ti4+ to Ti3+ and leading to the emergence of gap states, as monitored by in situ UV/X-ray photoelectron spectroscopy. The density functional theory calculation reveals that the water splitting becomes an exothermic process through hole oxidation with the assistance of the hydrogen-bond network. In addition to the widely concerned exotic activity from photocatalysts, our study demonstrates the internal hydrogen-bond network, which is ubiquitous at practical aqueous/catalyst interfaces, is also indispensable for water splitting.

VP-Detector: A 3D multi-scale dense convolutional neural network for macromolecule localization and classification in cryo-electron tomograms.

BACKGROUND AND OBJECTIVE: Cryo-electron tomography (cryo-ET) with subtomogram averaging (STA) is indispensable when studying macromolecule structures and functions in their native environments. Due to the low signal-to-noise ratio, the missing wedge artifacts in tomographic reconstructions, and multiple macromolecules of varied shapes and sizes, macromolecule localization and classification remain challenging. To tackle this bottleneck problem for structural determination by STA, we design an accurate macromolecule localization and classification method named voxelwise particle detector (VP-Detector). METHODS: VP-Detector is a two-stage particle detection method based on a 3D multiscale dense convolutional neural network (3D MSDNet). The proposed network uses 3D hybrid dilated convolution (3D HDC) to avoid the resolution loss caused by scaling operations. Meanwhile, it uses 3D dense connectivity to encourage the reuse of feature maps to reduce trainable parameters. In addition, the weighted focal loss is proposed to focus more attention on difficult samples and rare classes, which relieves the class imbalance caused by multiple particles of various sizes. The performance of VP-Detector is evaluated on both simulated and real-world tomograms, and it shows that VP-Detector outperforms state-of-the-art methods. RESULTS: The experiments show that VP-Detector outperforms the state-of-the-art methods on particle localization with an F1-score of 0.951 and a precision of 0.978. In addition, VP-Detector can replace manual particle picking in experiment on the real-world tomograms. Furthermore, it performs well in classifying large-, medium-, and small-weight proteins with accuracies of 1, 0.95, and 0.82, respectively. Finally, ablation studies demonstrate the effectiveness of 3D HDC, 3D dense connectivity, weighted focal loss, and training on small training sets. CONCLUSIONS: VP-Detector can achieve high accuracy in particle detection with few trainable parameters and support training on small datasets. It can also relieve the class imbalance caused by multiple particles with various shapes and sizes.

Ontology-aware deep learning enables ultrafast and interpretable source tracking among sub-million microbial community samples from hundreds of niches.

The taxonomic structure of microbial community sample is highly habitat-specific, making source tracking possible, allowing identification of the niches where samples originate. However, current methods face challenges when source tracking is scaled up. Here, we introduce a deep learning method based on the Ontology-aware Neural Network approach, ONN4MST, for large-scale source tracking. ONN4MST outperformed other methods with near-optimal accuracy when source tracking among 125,823 samples from 114 niches. ONN4MST also has a broad spectrum of applications. Overall, this study represents the first model-based method for source tracking among sub-million microbial community samples from hundreds of niches, with superior speed, accuracy, and interpretability. ONN4MST is available at https://github.com/HUST-NingKang-Lab/ONN4MST .

Time- and momentum-resolved image-potential states of 2H-MoS2 surface.

Rydberg-like image potential states (IPSs) form special series surface states on metal and semiconducting surfaces. Here, using time-resolved and momentum-resolved multi-photon photoemission (mPPE), we measured the energy positions, band dispersion, and carrier lifetimes of IPSs at the 2H-MoS2 surface. The energy minima of the IPSs (n = 1 and 2) were located at 0.77 and 0.21 eV below the vacuum level. In addition, the effective masses of these two IPSs are close to the rest mass of the free electron, clearly showing nearly-free-electron character. These properties suggest a good screening effect in the MoS2 parallel to the surface. The multi-photon resonances between the valence band and IPS (n = 1) are observed, showing a k‖-momentum-dependent behavior. Our time-resolved mPPE measurements show that the lifetime of photoexcited electrons in the IPS (n = 1) is about 33 fs.

DNN-DTIs: Improved drug-target interactions prediction using XGBoost feature selection and deep neural network.

Analysis and prediction of drug-target interactions (DTIs) play an important role in understanding drug mechanisms, as well as drug repositioning and design. Machine learning (ML)-based methods for DTIs prediction can mitigate the shortcomings of time-consuming and labor-intensive experimental approaches, while providing new ideas and insights for drug design. We propose a novel pipeline for predicting drug-target interactions, called DNN-DTIs. First, the target information is characterized by a number of features, namely, pseudo-amino acid composition, pseudo position-specific scoring matrix, conjoint triad composition, transition and distribution, Moreau-Broto autocorrelation, and structural features. The drug compounds are subsequently encoded using substructure fingerprints. Next, eXtreme gradient boosting (XGBoost) is used to determine the subset of non-redundant features of importance. The optimal balanced set of sample vectors is obtained by applying the synthetic minority oversampling technique (SMOTE). Finally, a DTIs predictor, DNN-DTIs, is developed based on a deep neural network (DNN) via a layer-by-layer learning scheme. Experimental results indicate that DNN-DTIs achieves better performance than other state-of-the-art predictors with ACC values of 98.78%, 98.60%, 97.98%, 98.24% and 98.00% on Enzyme, Ion Channels (IC), GPCR, Nuclear Receptors (NR) and Kuang's datasets. Therefore, the accurate prediction performance of DNN-DTIs makes it a favored choice for contributing to the study of DTIs, especially drug repositioning.

Formation of Ultra-High Temperature Ceramic Hollow Microspheres as Promising Lightweight Thermal Insulation Materials via a Molten Salt-Assisted Template Method.

Ultra-high temperature ceramics (UHTCs) have become a vital candidate material system for thermal protection systems in aerospace applications. However, high thermal conductivity and high density are the main obstacles to the application of UHTCs. It is a promising solution to prepare porous UHTCs using UHTC hollow microspheres (HMs) as a pore-forming agent. In this work, UHTC (ZrC, TiC, and HfC) HMs are successfully synthesized using carbon hollow microspheres (CHMs) as a template to react with metal powders in molten salt. The diameter of ZrC HMs is about 1 µm and the wall thickness is about 100 nm. The density of each microsphere and the volume fraction of ZrC are 3.36 g/cm3 and 48.42 vol %, respectively. The morphology, microstructure, and phase composition of the obtained ZrC HMs were characterized. The formation mechanism of the UHTC HMs was discussed. Porous ZrC ceramics were prepared using ZrC HMs as a pore-forming agent. The density and thermal conductivity of the porous ZrC ceramics are 3.12 g/cm3 and 1.82 W/(m·K), respectively, which are 53.64 and 91.12% lower than the density and thermal conductivity of dense ZrC ceramics, respectively. The results indicated that ZrC HMs are promising as pore-forming agents or a matrix for lightweight thermal insulation and high-temperature resistance applications in ultra-high temperature environments.

SVLR: Genome Structural Variant Detection Using Long-Read Sequencing Data.

Genome structural variants (SVs) have great impacts on human phenotype and diversity, and have been linked to numerous diseases. Long-read sequencing technologies arise to make it possible to find SVs of as long as 10,000 nucleotides. Thus, long read-based SV detection has been drawing attention of many recent research projects, and many tools have been developed for long reads to detect SVs recently. In this article, we present a new method, called SVLR, to detect SVs based on long-read sequencing data. Comparing with existing methods, SVLR can detect three new kinds of SVs: block replacements, block interchanges, and translocations. Although these new SVs are structurally more complicated, SVLR achieves accuracies that are comparable with those of the classic SVs. Moreover, for the classic SVs that can be detected by state-of-the-art methods (e.g., SVIM and Sniffles), our experiments demonstrate recall improvements of up to 38% without harming the precisions (i.e., >78%). We also point out three directions to further improve SV detection in the future. Source codes: https://github.com/GWYSDU/SVLR.

Determining structural and chemical heterogeneities of surface species at the single-bond limit.

The structure determination of surface species has long been a challenge because of their rich chemical heterogeneities. Modern tip-based microscopic techniques can resolve heterogeneities from their distinct electronic, geometric, and vibrational properties at the single-molecule level but with limited interpretation from each. Here, we combined scanning tunneling microscopy (STM), noncontact atomic force microscopy (AFM), and tip-enhanced Raman scattering (TERS) to characterize an assumed inactive system, pentacene on the Ag(110) surface. This enabled us to unambiguously correlate the structural and chemical heterogeneities of three pentacene-derivative species through specific carbon-hydrogen bond breaking. The joint STM-AFM-TERS strategy provides a comprehensive solution for determining chemical structures that are widely present in surface catalysis, on-surface synthesis, and two-dimensional materials.

Formation of Plasmonic Polarons in Highly Electron-Doped Anatase TiO2.

The existence of various quasiparticles of polarons because of electron-boson couplings plays important roles in determining electron transport in titanium dioxide (TiO2), which affects a wealth of physical properties from catalysis to interfacial superconductivity. In addition to the well-defined Fröhlich polarons whose electrons are dressed by the phonon clouds, it has been theoretically predicted that electrons can also couple to their own plasmonic oscillations, namely, the plasmonic polarons. Here we experimentally demonstrate the formation of plasmonic polarons in highly doped anatase TiO2 using angle-resolved photoemission spectroscopy. Our results show that the energy separation of plasmon-loss satellites follows a dependence on √n, where n is the electron density, manifesting the characteristic of plasmonic polarons. The spectral functions enable to quantitatively evaluate the strengths of electron-plasmon and electron-phonon couplings, respectively, providing an effective approach for characterizing the interplays among different bosonic modes in the complicate many-body interactions.

Creation of the Dirac Nodal Line by Extrinsic Symmetry Engineering.

The formation of the Dirac nodal line (DNL) requires intrinsic symmetry that can protect the degeneracy of continuous Dirac points in momentum space. Here, as an alternative approach, we propose an extrinsic symmetry protected DNL. On the basis of symmetry analysis and numerical calculations, we establish a general principle to design the nonsymmorphic symmetry protected 4-fold degenerate DNL against spin-orbit coupling in the nanopatterned 2D electron gas. Furthermore, on the basis of experimental measurements, we demonstrate the approximate realization of our proposal in the Bi/Cu(111) system, in which a highly dispersive DNL is observed at the boundary of the Brillouin zone. We envision that the extrinsic symmetry engineering will greatly enhance the ability for artificially constructing the exotic topological bands in the future.

Dilated-DenseNet For Macromolecule Classification In Cryo-electron Tomography.

Cryo-electron tomography (cryo-ET) combined with subtomogram averaging (STA) is a unique technique in revealing macromolecule structures in their near-native state. However, due to the macromolecular structural heterogeneity, low signal-to-noise-ratio (SNR) and anisotropic resolution in the tomogram, macromolecule classification, a critical step of STA, remains a great challenge. In this paper, we propose a novel convolution neural network, named 3D-Dilated-DenseNet, to improve the performance of macromolecule classification in STA. The proposed 3D-Dilated-DenseNet is challenged by the synthetic dataset in the SHREC contest and the experimental dataset, and compared with the SHREC-CNN (the state-of-the-art CNN model in the SHREC contest) and the baseline 3D-DenseNet. The results showed that 3D-Dilated-DenseNet significantly outperformed 3D-DenseNet but 3D-DenseNet is well above SHREC-CNN. Moreover, in order to further demonstrate the validity of dilated convolution in the classification task, we visualized the feature map of 3D-Dilated-DenseNet and 3D-DenseNet. Dilated convolution extracts a much more representative feature map.